Neural-based learning in grasp force control of a robot hand

In this paper, we present a new control system for the intelligent force control of multifingered robot grips which combines both fuzzy-based adaptation level and a neural-based one with a conventional PID-controller. The most attention is given to the neural-based force adaptation level implemented by three-layered back-propagation neural networks. A computer based simulation system for the peg-in-hole insertion task is developed to analyze the capabilities of the neural controllers. Their behaviour is discussed by comparing them to conventional and fuzzy-based force controllers performing the same task.     

Generation of an optimal architecture of neuro force controllers for robot manipulators in unknown environments using genetic programming with fuzzy fitness evaluation

 In this paper, we have applied genetic programming to generate an optimal architecture of neuro force controllers for robot manipulators in any environment. In order to perform precise force control in unknown environments, the optimal structured neuro force controller is generated using genetic programming with fuzzy fitness evaluation. After the architecture of the neuro controller has been optimized for any kinds of environments, it can be applied for a robot contact task with an unknown environment in on-line manner using its own adaptation ability. An effective crossover operation is proposed for the efficient evolution of the controllers. The simulation has been carried out to evaluate the effectiveness of the proposed robot force controller.     

Humanoid robot arm performance optimization using multi objective evolutionary algorithm

As humanoid robots are expected to operate in human environments they are expected to perform a wide range of tasks. Therefore, the robot arm motion must be generated based on the specific task. In this paper we propose an optimal arm motion generation satisfying multiple criteria. In our method, we evolved neural controllers that generate the humanoid robot arm motion satisfying three different criteria; minimum time, minimum distance and minimum acceleration. The robot hand is required to move from the initial to the final goal position. In order to compare the performance, single objective GA is also considered as an optimization tool. Selected neural controllers from the Pareto solution are implemented and their performance is evaluated. Experimental investigation shows that the evolved neural controllers performed well in the real hardware of the mobile humanoid robot platform.     

Bio-inspired behaviour-based control

The design and development of conventional controllers for robot platforms are sometimes too complex to achieve due to the fact that they require an exact model of the system and of the operating environment. The ability to pre-account for unknown operating environments is an important task for the controller to be robust. In contrast, biological controllers are model free and are based on simple working principles. Due to natural biological principles these controllers are adaptive and more robust than their conventional counterparts. In this paper, a behaviour-based controller has been developed, inspired by the concept of spinal fields found in frogs and rats. The performance of the controller has been verified on a Khepera robot platform.     

基于迭代学习的机械手操作空间力/位置混合控制算法

基于对常规机械手操作空间力/位置混合控制算法的简单回顾,及对该算法所遇到困难的分析,提出了一种基于迭代学习的机械手操作空间力/位置混合控制算法,来改善机械手同高刚度环境接触时,机械手力/位置混合控制的动态控制性能．给出了学习算法的收敛条件及其证明.实验表明该算法具有快速的收敛性,能达到很高的力/位置动态控制精度.     

On the decoupling of position and force controllers in constrained robotic tasks

In hybrid control of robot manipulators separate controllers are designed for force and position errors control. Controllers are designed either in task or joint space and their outputs combine to provide input torque to the manipulator. Position and force controllers performance in a constrained robotic task is affected by their interaction to a degree dependent on the controller's ability to reject disturbances. Ideally, decoupling of the two control loops is desired to achieve the best performance in position and force directions. In this article, analysis of control loop interactions is performed for contact and noncontact phases, and controller design requirements are developed to achieve maximum decoupling. Design requirements involve output subspace of each controller leading to control discontinuities for contact and noncontact phases. In the noncontact phase, satisfaction of design requirements leads to a fully linearized and decoupled system. When in contact with the constraining surface, design requirements eliminate disturbances in the force loop, but minimize disturbances in the position loop to an extent dependent on force loop performance. Known hybrid control schemes analysis is performed to reveal existence of control loop interactions in these schemes. Confirmation of theoretical analysis is done through simulation of a three revolute planar manipulator. © 1998 John Wiley & Sons, Inc.     

基于迭代学习的机械手操作空间力/位置混合控制算法

基于对常规机械手操作空间力／位置混合控制算法的简单回顾，及对该算法所遇到困难的分析，提出了一种基于迭代学习的机械手操作空间力／位置混合控制算法，来改善机械手同高刚度环境接触时，机械手力／位置混合控制的动态控制性能．给出了学习算法的收敛条件及其证明．实验表明该算法具有快速的收敛性，能达到很高的力／位置动态控制精度．     

Experiments in contact control

This article describes the implementation, experimentation, and application of contact control schemes for a 7-DOF Robotics Research arm. The contact forces and torques are measured in the sensor frame by the 6-axis force/torque sensor mounted at the wrist, are compensated for gravity, and then are transformed to the tool frame in which the contact task is defined and executed. The contact control schemes are implemented on the existing robot Cartesian position control system at 400Hz, do not require force rate information, and are extremely simple and computationally fast. Three types of contact control schemes are presented: compliance control, force control, and dual-mode control. In the compliance control scheme, the contact force is fed back through a lag-plus-feedforward compliance controller so that the end-effector behaves like a spring with adjustable stiffness; thus the contact force can be controlled by the reference position command. In the force control scheme, a force setpoint is used as the command input and a proportional-plus-integral force controller is employed to ensure that the contact force tracks the force setpoint accurately. In the dual-mode control scheme, the end-effector approaches and impacts the reaction surface in compliance mode, and the control scheme is then switched automatically to force mode after the initial contact has been established. Experimental results are presented to demonstrate contact with hard and soft surfaces under the three proposed control schemes. The article is concluded with the application of the proposed schemes to perform a contact-based eddy-current inspection task. In this task, the robot first approaches the inspection surface in compliance control until it feels that it has touched the surface, and then automatically levels the end-effector on the surface. The robot control system then transitions to force control and applies the desired force on the surface while executing a scanning motion. At the completion of the inspection task, the robot first relaxes the applied force and then retracts from the surface. © 1996 John Wiley & Sons, Inc.     

仿人机器人控制系统的研究与实现

根据仿人机器人控制性能的要求,设计开发了关节控制器,并通过CAN总线把各个关节控制器、力传感器及上位机连接在一起,构成了分布式控制系统．利用无线局域网技术,实现了语音、视频等多媒体信息的传输,把监控台、头部、上身和移动平台连接在一起,构成了仿人机器人完整的控制系统．最后提出了一些设想以提高系统的性能．     

Robotic assembly automation using robust compliant control

Industrial robots used to perform assembly applications are still a small portion of total robot sales each year. One of the main reasons is that it is difficult for conventional industrial robots to adapt to any sort of change. This paper proposes a robust control strategy to perform an assembly task of inserting a printed circuit board (PCB) into an edge connector socket using a SCARA robot. The task is very challenging because it involves compliant manipulation in which a substantial force is needed to accomplish the insertion operation and there are some dynamic constraints from the environment. Therefore, a robust control algorithm is developed and used to perform the assembly process. The dynamic model of the robotic system is developed and the dynamic parameters are identified. Experiments were performed to validate the proposed method. Experimental results show that the robust control algorithm can deal with parameter uncertainties in the dynamic model, thus achieve better performance than the model based control method. An abnormal case is also investigated to demonstrate that the robust compliant control method can deal with the abnormal situation without damaging the system and assembly parts, while pure position control method may cause damages. This strategy can also be used in other similar assembly processes with compliant applications.     

Adaptive force and position control of manipulators

The article presents simple methods for the design of adaptive force and position controllers for robot manipulators within the hybrid control architecture. The force controller is composed of an adaptive PID feedback controller, an auxiliary signal, and a force feedforward term, and achieves tracking of desired force setpoints in the constraint directions. The position controller consists of adaptive feedback and feedforward controllers as well as an auxiliary signal, and accomplishes tracking of desired position trajectories in the free directions. The controllers are capable of compensating for dynamic cross-couplings that exist between the position and force control loops in the hybrid control architecture. The adaptive controllers do not require knowledge of the complex dynamic model or parameter values of the manipulator or the environment. The proposed control schemes are computationally fast and suitable for implementation in online control with high sampling rates. The methods are applied to a two-link manipulator for simultaneous force and position control. Simulation results confirm that the adaptive controllers perform remarkably well under different conditions.     

Multiobjective Evolution of Neural Controllers and Task Complexity

Robots operating in everyday life environments are often required to switch between different tasks. While learning and evolution have been effectively applied to single task performance, multiple task performance still lacks methods that have been demonstrated to be both reliable and efficient. This paper introduces a new method for multiple task performance based on multiobjective evolutionary algorithms, where each task is considered as a separate objective function. In order to verify the effectiveness, the proposed method is applied to evolve neural controllers for the Cyber Rodent (CR) robot that has to switch properly between two distinctly different tasks: 1) protecting another moving robot by following it closely and 2) collecting objects scattered in the environment. Furthermore, the tasks and neural complexity are analyzed by including the neural structure as a separate objective function. The simulation and experimental results using the CR robot show that the multiobjective-based evolutionary method can be applied effectively for generating neural networks that enable the robot to perform multiple tasks simultaneously.     

Evolution of Fuzzy Controllers for Robotic Vehicles: The Role of Fitness Function Selection

An important issue not addressed in the literature, is related to the selection of the fitness function parameters which are used in the evolution process of fuzzy logic controllers for mobile robot navigation. The majority of the fitness functions used for controllers evolution are empirically selected and (most of times) task specified. This results to controllers which heavily depend on fitness function selection. In this paper we compare three major different types of fitness functions and how they affect the navigation performance of a fuzzy logic controlled real robot. Genetic algorithms are employed to evolve the membership functions of these controllers. Further, an efficiency measure is introduced for the systematic analysis and benchmarking of overall performance. This measure takes into account important performance results of the robot during experimentation, such as the final distance from target, the time needed to reach its final position, the time of sensor activation, the mean linear velocity e.t.c. In order to examine the validity of our approach a low cost mobile robot has been developed, which is used as a testbed.     

Efficiency evaluation of robots in machining applications using industrial performance measure

The paper is devoted to the robotic based machining. The main focus is made on robot accuracy in milling operation and evaluation robot capacity to perform the task with desired precision. Particular attention is paid to the proper modeling of manipulator stiffness properties and the cutting force estimation. In contrast to other works, the robot performance is evaluated using the circularity norm that evaluates the contortion degree of the benchmark circle to be machined. The developed approach is applied to five industrial robots of KUKA family, which have been ranked for several machining tasks. The validity of the proposed technique was confirmed by experimental study dealing with robot-based machining of circular grooves for several workpiece samples and different locations.     

Control of bidirectional physical human–robot interaction based on the human intention

This paper presents a control strategy for human–robot interaction with physical contact, recognizing the human intention to control the movement of a non-holonomic mobile robot. The human intention is modeled by mechanical impedance, sensing the human-desired force intensity and the human-desired force direction to guide the robot through unstructured environments. Robot dynamics is included to improve the interaction performance. Stability analysis of the proposed control system is proved by using Lyapunov theory. Real experiments of the human–robot interaction show the performance of the proposed controllers.     

自适应模糊与CMAC并行的机器人力/位置控制

为提高机器人系统对机器人末端操纵器与外界工作环境接触时,其接触刚度不确定性的自适应能力,在机器人力／位置混合控制的基础上,设计出了一种基于自适应模糊与CMAC并行控制的机器人力控制器,采用小脑模型神经控制器实现前馈控制,实现被控对象的逆动态模型,自适应模糊控制器实现反馈控制,保证系统的稳定性,且抑制扰动。以平面两关节机器人进行仿真,仿真结果表明,系统的自适应能力和力跟踪能力有显著的提高,机械手在其末端操纵器与刚性变化范围较大的外界工作环境接触时,具有较强的适应能力,较好地完成了机器人的力／位置控制。     

Robot Control Optimization Using Reinforcement Learning

Conventional robot control schemes are basically model-based methods. However, exact modeling of robot dynamics poses considerable problems and faces various uncertainties in task execution. This paper proposes a reinforcement learning control approach for overcoming such drawbacks. An artificial neural network (ANN) serves as the learning structure, and an applied stochastic real-valued (SRV) unit as the learning method. Initially, force tracking control of a two-link robot arm is simulated to verify the control design. The simulation results confirm that even without information related to the robot dynamic model and environment states, operation rules for simultaneous controlling force and velocity are achievable by repetitive exploration. Hitherto, however, an acceptable performance has demanded many learning iterations and the learning speed proved too slow for practical applications. The approach herein, therefore, improves the tracking performance by combining a conventional controller with a reinforcement learning strategy. Experimental results demonstrate improved trajectory tracking performance of a two-link direct-drive robot manipulator using the proposed method.     

Robust Motion Control for Robotic Assembly Tasks using Variable Structure Control Scheme

The success of robot assembly tasks depends heavily on its ability to handle the interactions which take place between the parts being assembled. In this paper, a robust motion-control method is presented for robot manipulators performing assembly tasks in the presence of dynamic constraints from the environment. Using variable structure model reaching control concept, the control objectives is first formulated as a performance model in the task space. A dynamic compensator is then introduced to form the switching function such that the sliding-mode matches the desired model. A simple variable structure control law is suggested to force the system to reach and stay on the sliding mode so that the specified model is achieved.The proposed method is applied to control the prismatic joint of a selective compliance assembly robot-arm type robot for the insertion of printed circuit board into an edge connector socket. Various amounts of interaction forces are generated during the operation. Experimental and simulation results demonstrated the performance of the variable structure model reaching control approach. In comparison, it is shown that the popular position controllers such as proportional plus derivative control and proportional plus derivative with model-based feedforward control are not suitable for achieving good trajectory tracking accuracy in assembly tasks which experience potential interaction force.     

Evolving reliable and robust controllers for real robots by genetic programming

 Using Genetic Programming (GP)-based approaches to evolve robot controllers has the advantage of operating variable-size genotype. This is an important feature for evolving robot control systems as it allows complete freedom for the control architecture in respect to the task complexity which is difficult to predict. However, GP-based work in evolving controllers has been questioned in the verification of the performance on real robots, the generalisation of defining primitives, and the computational cost needed. In this paper, we present our GP framework in which a special representation of the robot controller is designed; this representation can capture well the characteristic of a behaviour controller so that our system can efficiently evolve desired robot behaviours by a relatively low computational cost. This system has been successfully used to evolve reliable and robust controllers working on a real robot, for a variety of tasks.     

Fuzzy control-based real-time robust balance for a humanoid robot

This paper studies and implements a real-time robust balance control for a humanoid robot under three environment disturbances which are an external thrust, an inclinable platform, and a see-saw. More precisely to say, the robot with robust control can resist an external thrust, stand on a two-axis inclinable platform, or walk on a see-saw successfully. The main feature of the robot is that it has a waist joint which has three degrees of freedom. With the aids of the proposed fuzzy controllers, the robot can change the posture of the body nimbly by adjusting the waist joint and two ankle joints to strengthen the stabilization capacity. The sensory system of the robot includes eight force sensors and one inertial measurement unit sensor in order to measure the center of pressure and the slant angle of the robot’s body. According to the measured data from the sensors and by imitating human reflex actions, the proposed fuzzy controllers perform real-time balance control for the robot under three environment disturbances. According to the experiment results, the stability of the robot is increased at least 32.2 and 61.7% under the first two environment disturbances, respectively. In addition, the robot walking on a see-saw has a success rate of about 95%.